Andrew Chernow <pg-job@esilo.com> writes:
> I wasn't saying to do this each time you run a backup, geez that would be
> horrible. Pickup from where you left off the last time you backed up
> data/records. How many images and how much data is being generated in a 60
> second period? I dought 3 billion files and hundreds of terabytes. When you
> know what your data generation is, you know what resources you need to
> replicate this information to a backup server (local or remote).
I'm not talking about backups. I'm talking about restores.
> How is this any different than db replication. It would have to backup the
> same amount of information? You would require the same horse power and
> bandwidth.
The difference is that I'd restore the data and then sync the difference from
some point in time to "now". The referential integrity would be guaranteed by
the database itself and I won't have any pointers to files that doesn't exist
or files without pointers to it.
--
Jorge Godoy <jgodoy@gmail.com>